What to Keep and What to Drop: Adaptive Table Filtering Framework

TL;DR

The paper introduces ATF, an adaptive, question-aware table filtering framework that reduces table size by 70% to improve large language model reasoning on tables, especially for out-of-domain tasks.

Contribution

ATF is a modular filtering pipeline that uses LLM-generated descriptions and clustering to prune tables without retraining models, enhancing reasoning efficiency.

Findings

Reduces table size by 70%

Improves out-of-domain TableQA performance

Slightly decreases performance on Table Fact Verification

Abstract

Large language models (LLMs) for table-based reasoning often struggle with large tables due to input length limits. We propose ATF (Adaptive Table Filtering Framework), a modular and question-aware filtering pipeline that prunes uninformative columns and rows using LLM-generated column descriptions, clustering, and sparse-dense alignment scores. ATF integrates seamlessly with existing models (e.g., TAPAS, TAPEX) without retraining. Experiments show that ATF reduces table cells by 70%, boosting performance on out-of-domain TableQA tasks while causing slight performance drops on Table Fact Verification, where full-table context is more critical. These results highlight ATF's ability to adaptively balance informativeness and minimalism across tasks. Our code available at: https://github.com/torijune/ATF-Adaptive-Table-Filtering-Framework